[comp.sys.sgi] building X11R4 on personal iris

tuna@ai.mit.edu (Kirk 'UhOh' Johnson) (07/03/90)

the research group i work with has recently aquired a personal iris.
unfortunately, we discovered that the X11 stuff that comes with the
standard distribution (irix 3.2.1) includes only the server and client
binaries. none of the libraries or include files necessary to build
X11-based applications are present.

according to our sales rep, the include files and libraries are
available from SGI, but at an extra cost of about $1000 (less some
academic discount).

while $1000 isn't a huge sum of money, it strikes me that i _should_
be able to get the "raw" X11R4 sources via ftp and simply build all
of the non-server stuff. (the fact that the X11R4 sources provide an
"sgi.cf" file in mit/config fuels this gut feeling further.)

unfortunately, i can't seem to build successfully. i've checked that
OSMajorVersion and OSMinorVersion in mit/config/sgi.cf are correct (3
and 2, respectively). if i add the line

    #define BuildServer    NO

to the end of sgi.cf (so that the make doesn't try to build a server)
and then fire up a make World, the compilation dies at line 69 of
mit/util/makedepend/main.c. lines 68 through 81 follow:

    #ifndef USG
    struct sigvec sig_vec = {
            catch,
             (1<<(SIGINT -1))
            |(1<<(SIGQUIT-1))
            |(1<<(SIGBUS-1))
            |(1<<(SIGILL-1))
            |(1<<(SIGSEGV-1))
            |(1<<(SIGHUP-1))
            |(1<<(SIGPIPE-1))
            |(1<<(SIGSYS-1)),
            0
    };
    #endif /* USG */
    
the problem, complains the compiler, is that struct sigvec is not
defined. since i was unable to find a definition in any of the include
files i looked in (and taking a hint from some code further down in
mit/util/makedepend/main.c), i wondered if the problem was that USG
wasn't #define-d. 

to finish up a long story, adding "-DUSG" to the StandardDefines in
mit/config/sgi.cf didn't help much. although make World got quite a
bit further, it did crash (fail to build successfully, that is)
eventually. i could provide more details about the specific problems
this build ran into, but i'm not sure it'd really be of significant
utility.

ok. now for the real intent of my posting. 

(1) am i off my rocker to suspect that i should be able to get the
    missing include files and libraries by compiling the X11R4 source
    (with "#define BuildServer NO")? (sure, this would leave me with
    X11R4 clients and an X11R3 server, but that's not a problem.)

(2) if i'm not off my rocker, does anybody know what incantations
    required for a successful X11R4 build?

any help would be greatly appreciated. please followup via e-mail,
i'll post a summary if there is sufficient interest.

thanks

kirk
--

----------------------------------------------------------------------
kirk johnson                                           `Eat blue dogs
tuna@masala.lcs.mit.edu                                 and dig life.'

jim@baroque.Stanford.EDU (James Helman) (07/04/90)

Back in April I posted the patches I used to compile the X11R4
distribution on our 4D/220 under IRIX 3.2 (without server, of course).
The patches are available by anonymous ftp from fresnel.stanford.edu
in pub/4DX.  In addition to the patches, the X11R4 client binaries,
includes and libraries are also there.  Caveat 1: I haven't tried
recompiling under 3.3 yet.  Caveat 2: XPutImage in SGI's X server
(IRIX 3.2 and 3.3) has several problems drawing 1 bit images, e.g.
using xwud to display window dumps from a monochrome Sun doesn't work.

Jim Helman
Department of Applied Physics			6 Trillium Lane
Stanford University				San Carlos, CA 94070
(jim@thrush.stanford.edu) 			(415) 723-9127